Sample selection by diagnosis in clinical drug evaluations.

摘要: Theoretical models for sample selection by diagnosis using one, one-out-of-two, two, two-out-of-three, and three concurring opinions are illustrated. These then applied to two categories, schizophrenia depressive disorders, an actual of patients diagnosed multiple observers. The conclusion is reached that a single opinion not sufficiently reliable research studies. choice alternative methods depends on the needs particular study. Samples chosen basis two-out-of-three provide least total error. When erroneous inclusions particularly undesirable, higher degrees consensus—e.g. agreeing pairs or triplets—are necessary.